recker 1.0.79 → 1.0.80-next.cdb4529

package/dist/seo/analyzer.js

@@ -39,6 +39,7 @@ export class SeoAnalyzer {
         const socialLinkDetails = this.analyzeSocialLinks(links, socialDomains);
         const headings = this.analyzeHeadings();
         const content = this.analyzeContent(headings);
+        const contentSections = this.extractContentSectionSignals(headings);
         const linkAnalysis = this.buildLinkAnalysis(links);
         const imageAnalysis = this.buildImageAnalysis(images);
         const social = this.buildSocialAnalysis(og, twitter);
@@ -48,6 +49,16 @@ export class SeoAnalyzer {
         const feeds = this.analyzeFeeds();
         const conversion = this.analyzeConversionElements(links, visibleText);
         const pageType = this.detectPageType(jsonLd);
+        const linkAnchorTexts = links
+            .map((link) => link.text?.trim())
+            .filter((text) => Boolean(text))
+            .filter((text) => text.length <= 140)
+            .filter((text) => !text.match(/^\s*$/))
+            .slice(0, 80);
+        const linkUrlSamples = links
+            .map((link) => extractReadablePathFromUrl(link.href))
+            .filter((sample) => sample.length > 0)
+            .slice(0, 80);
         const context = this.buildRuleContext({
             pageType,
             meta,
@@ -124,12 +135,15 @@ export class SeoAnalyzer {
                 items: jsonLd,
             },
             headings: headings,
+            contentSections,
             content,
             keywords,
             links: linkAnalysis,
             images: imageAnalysis,
             social,
             technical,
+            linkAnchorTexts,
+            linkUrlSamples,
         };
     }
     getMainBody() {
@@ -181,6 +195,97 @@ export class SeoAnalyzer {
             return 'other';
         }
     }
+    extractContentSectionSignals(headings) {
+        const body = this.getMainBody();
+        if (!body)
+            return [];
+        const candidateTags = 'h1,h2,h3,h4,h5,h6,p,li,ol,ul,article,section,main,figure,figcaption,blockquote,table,th,td';
+        const nodes = body.querySelectorAll(candidateTags);
+        const normalizeSignalText = (value) => value
+            .replace(/\s+/g, ' ')
+            .replace(/[,\.;:!?()[\]{}"“”']/g, '')
+            .trim()
+            .toLowerCase();
+        const headingStack = [];
+        const seen = new Set();
+        const results = [];
+        const pushSectionSignal = (entry) => {
+            const fingerprint = `${entry.tagName}|${entry.headingPath.join(' > ')}|${entry.text.slice(0, 120)}`;
+            if (seen.has(fingerprint))
+                return;
+            seen.add(fingerprint);
+            results.push(entry);
+        };
+        for (const node of nodes) {
+            const tagName = node.tagName.toLowerCase();
+            if (/^h[1-6]$/.test(tagName)) {
+                const level = Number.parseInt(tagName.slice(1), 10);
+                const text = normalizeSignalText(node.text);
+                if (!text || text.length < 4) {
+                    continue;
+                }
+                while (headingStack.length > 0 && headingStack[headingStack.length - 1].level >= level) {
+                    headingStack.pop();
+                }
+                headingStack.push({ level, text });
+                const headingPath = headingStack.map((entry) => entry.text);
+                const source = {
+                    headingPath,
+                    source: 'heading',
+                    tagName,
+                    headingLevel: level,
+                    text,
+                    wordCount: text.split(/\s+/).length,
+                    linkDensity: 0,
+                    weight: 1.45,
+                };
+                pushSectionSignal(source);
+                continue;
+            }
+            if (tagName === 'ul' || tagName === 'ol') {
+                continue;
+            }
+            const text = normalizeSignalText(node.text);
+            if (!text || text.split(/\s+/).length < 4) {
+                continue;
+            }
+            const words = text.split(/\s+/);
+            const wordCount = words.length;
+            if (wordCount < 4)
+                continue;
+            const anchorCount = node.querySelectorAll('a').length;
+            const linkDensity = wordCount > 0
+                ? Math.min(1, anchorCount / Math.max(1, wordCount))
+                : 0;
+            if (linkDensity > 0.85)
+                continue;
+            const headingPath = headingStack.length > 0
+                ? headingStack.map((entry) => entry.text)
+                : headings.h1Count > 0
+                    ? ['conteúdo']
+                    : ['home'];
+            const sourceLabel = tagName === 'li' ? 'list-item'
+                : tagName === 'figure' || tagName === 'figcaption' ? 'figure'
+                    : (tagName === 'table' || tagName === 'th' || tagName === 'td') ? 'table'
+                        : 'paragraph';
+            const headingLevel = headingStack.length > 0 ? headingStack[headingStack.length - 1].level : 1;
+            const baseWeight = sourceLabel === 'paragraph' ? 1.25 : 1.05;
+            const adjustedWeight = baseWeight * (1 - (linkDensity * 0.4));
+            pushSectionSignal({
+                headingPath,
+                source: sourceLabel,
+                tagName,
+                headingLevel,
+                text,
+                wordCount,
+                linkDensity,
+                weight: Math.max(0.5, adjustedWeight),
+            });
+        }
+        return results
+            .filter((entry) => entry.text.length >= 20)
+            .slice(0, 160);
+    }
     getVisibleText() {
         const body = this.getMainBody();
         if (!body)
@@ -1221,7 +1326,7 @@ export class SeoAnalyzer {
                 const level = parseInt(tagName.substring(1), 10);
                 const text = el.text.trim();
                 counts[level] = (counts[level] || 0) + 1;
-                structure.push({ level, text: text.slice(0, 80), count: 1 });
+                structure.push({ level, text: text.slice(0, 120), count: 1 });
                 if (level === 2) {
                     if (inSection) {
                         sectionWordCounts.push(currentSectionWordCount);
@@ -1261,8 +1366,24 @@ export class SeoAnalyzer {
         else if (counts[1] > 1) {
             issues.push('Multiple H1 tags');
         }
+        const tree = [];
+        const stack = [];
+        for (const heading of structure) {
+            const node = { level: heading.level, text: heading.text, children: [] };
+            while (stack.length > 0 && stack[stack.length - 1].level >= heading.level) {
+                stack.pop();
+            }
+            if (stack.length === 0) {
+                tree.push(node);
+            }
+            else {
+                stack[stack.length - 1].node.children.push(node);
+            }
+            stack.push({ node, level: heading.level });
+        }
         return {
             structure,
+            tree,
             h1Count: counts[1],
             hasProperHierarchy,
             issues,
@@ -1495,6 +1616,28 @@ export class SeoAnalyzer {
         return this.rulesEngine.getCategories();
     }
 }
+function extractReadablePathFromUrl(value) {
+    try {
+        const parsed = new URL(value);
+        const fragments = parsed.pathname
+            .split('/')
+            .filter((segment) => segment.length > 0)
+            .slice(0, 3)
+            .map((segment) => segment.replace(/[-_]/g, ' '))
+            .join(' ')
+            .replace(/\d+/g, ' ')
+            .replace(/\s+/g, ' ')
+            .trim();
+        const host = parsed.hostname.replace(/^www\./, '').replace(/\./g, ' ');
+        if (fragments.length > 0) {
+            return `${host} ${fragments}`.trim();
+        }
+        return host;
+    }
+    catch {
+        return '';
+    }
+}
 export async function analyzeSeo(html, options = {}) {
     const analyzer = await SeoAnalyzer.fromHtml(html, options);
     return analyzer.analyze();

package/dist/seo/index.d.ts

@@ -6,7 +6,7 @@ export type { SeoSpiderOptions, SeoPageResult, SiteWideIssue, SeoSpiderResult, }
 export { SeoRulesEngine, createRulesEngine, SEO_THRESHOLDS, ALL_SEO_RULES, } from './rules/index.js';
 export { generateSeoFilename, resolveOutputPath, writeReport, formatReportForJson, } from './output.js';
 export type { SeoOutputType, OutputOptions, WriteOptions } from './output.js';
-export type { SeoReport, SeoCheckResult, SeoStatus, SeoPageType, SeoTiming, HeadingAnalysis, HeadingInfo, ContentMetrics, LinkAnalysis, ImageAnalysis, SocialMetaAnalysis, TechnicalSeo, SeoAnalyzerOptions, } from './types.js';
+export type { SeoReport, SeoCheckResult, SeoStatus, SeoPageType, SeoTiming, HeadingAnalysis, HeadingInfo, HeadingNode, ContentMetrics, LinkAnalysis, ImageAnalysis, SocialMetaAnalysis, TechnicalSeo, SeoAnalyzerOptions, } from './types.js';
 export type { SeoRule, RuleContext, RuleResult, RuleEvidence, RuleCategory, RuleSeverity, RulesEngineOptions, } from './rules/index.js';
 export type { SeoAnalyzerFullOptions } from './analyzer.js';
 export { parseRobotsTxt, validateRobotsTxt, isPathAllowed, fetchAndValidateRobotsTxt, } from './validators/robots.js';

package/dist/seo/keyword-campaign-analyzer.d.ts

@@ -0,0 +1,2 @@
+import { type KeywordCampaignOptions, type KeywordCampaignReport } from './keyword-campaign-shared.js';
+export declare function analyzeKeywordCampaign(options: KeywordCampaignOptions): Promise<KeywordCampaignReport>;